The River of AI and Machine Learning
When did machines move from following rules to learning patterns from data?
From Boole to ChatGPT — 170 years toward machines that learn.
Read the current
What changed between the cities?
- 1
1854–1937
scholars and schools
LincolnCorkBoole — Logic as Arithmetic
Boole reduced logic to binary-style algebra in The Laws of Thought at Queen’s College Cork. In 1937, Shannon showed at MIT how Boolean algebra could describe switching circuits.
Boolean algebra in Cork and Shannon’s 1937 MIT master’s thesisPlace this segment on the map - 2
1950–1958
scholars and schools
ManchesterPrincetonTuring — Can Machines Think?
While at Manchester, Turing proposed an operational test for machine intelligence in his 1950 paper “Computing Machinery and Intelligence.” The Dartmouth workshop named artificial intelligence as a field in 1956.
Computing Machinery and Intelligence (1950, Manchester) and the Dartmouth workshop (1956)Place this segment on the map - 3
1958–1986
scholars and schools
BuffaloTorontoRosenblatt and Backprop — Foundations of Neural Networks
Frank Rosenblatt presented the Mark I perceptron at the Cornell Aeronautical Laboratory in Buffalo in 1958. Work by Werbos and then Rumelhart, Hinton, and Williams helped establish backpropagation by 1986.
The perceptron (1958, CAL Buffalo), backpropagation (1986, UCSD/CMU/Toronto), and universal approximationPlace this segment on the map - 4
2012–2024
scholars and schools
TorontoMountain ViewDeep Learning Revolution
AlexNet’s 2012 ImageNet victory in Toronto was followed by the Transformer at Google Brain in 2017 and ChatGPT at OpenAI in 2022—a decade of rapid change.
AlexNet (2012, Toronto), Transformer (2017, Google Brain), and ChatGPT (2022, OpenAI San Francisco)Place this segment on the map
How to read the lines
Each line is an editorial route through problems, texts, and practices. It does not imply one book moving in a straight line, a sole invention, or identical adoption everywhere.